| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
| |
os.path.join barfs if we're passing it 0 arguments, as would happen
when doing "import foo".
|
|
|
|
|
|
|
|
|
| |
Rather than introspecting the current directory, use the environment
variable we know is present if and only if we're running uninstalled.
This will be less prone to failure when we switch to nonrecursive
make, since the generated script will be in the top builddir, not
tools/ anymore.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The C compiler will pick an enumeration type that accomodates the specified
values for the enumeration, so ignoring 64-bit enumerations, we can
have enumeration values from MININT32 to MAXUINT32. To handle this properly:
- Use gint64 for holding eumeration values when scanning
- Add a 'unsigned_value' bit to ValueBlob so we can distinguish the
int32 vs. uint32 cases in the typelib
- Change the return value of g_value_info_get_value() to gint64.
https://bugzilla.gnome.org/show_bug.cgi?id=629704
|
|
|
|
|
|
|
|
|
|
| |
Previously we just were sloppy and didn't bother to accurately compute
signed/unsigned for enumeration types. But since we expect bindings
to decode a field value or function return value from an integer to
an enumeration they have know whether an integer value is
0xffffffff or -1, so we need to do the full computation.
https://bugzilla.gnome.org/show_bug.cgi?id=629704
|
|
|
|
| |
This is a preparatory patch for adding a new basic tag.
|
|
|
|
|
|
| |
Otherwise people may not know about --warn-all.
https://bugzilla.gnome.org/show_bug.cgi?id=631644
|
| |
|
|
|
|
|
| |
The compiler path explicitly split()s the variable, we should do
so in the linker path too.
|
|
|
|
|
|
|
| |
Commit f97cc8687469f25752f79275 broke the lookup in
g_irepository_get_info; the passed offset is 0-based, then
we convert it to 1-based (and then back to 0 later...which needs
to be fixed).
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Before, we were adding the --libs-only-L from pkg-config files
during initial scanning, as if --library-path was specified,
*and* inside the dumper, we used --libs.
Fix this by always doing -L. <introspected libraries> <pkg-config libs>
This should ensure we don't duplicate the search path anyways.
https://bugzilla.gnome.org/show_bug.cgi?id=631348
|
|
|
|
|
|
| |
Might as well be liberal here, as long as we have a test case.
https://bugzilla.gnome.org/show_bug.cgi?id=631690
|
|
|
|
|
|
|
|
|
|
|
| |
This is a cleanup patch in preparation for future indexing
patches.
The lookup code was a mess trying to mash in the 3 different
cases of name, GType, and index into one mega-function.
Split it up properly, and move the core typelib internal-scanning
bits into gitypelib.c where it belongs.
|
| |
|
|
|
|
|
|
| |
Previously we bombed out if the minor version didn't match what
we expected; this was silly since the whole point of the minor
version is to enumerate compatible changes.
|
|
|
|
|
| |
...by default. However, the GI_SCANNER_DEBUG=save-temps option
helps for debugging.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
OS vendors which compile from the part-binary "tarballs" which
include e.g. pregenerated gtk-doc, they typically don't pass
--enable-gtk-doc. So conditionalizing "make install"
for docs on ENABLE_GTK_DOC won't work.
This is the same as GTK+ and other projects currently work.
|
|
|
|
|
| |
I don't have a GPG key setup anymore, not interested in
the archaeology to find it right now.
|
| |
|
|
|
|
| |
This reverts commit d15e386c8e10dd6b645a444e29cc053100089a07.
|
|
|
|
| |
This reverts commit 754e5462f6dfbe3793eb959abf3dbfa7d2fe6435.
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=606686
|
| |
|
|
|
|
| |
Previously, such annotations would be ignored.
|
| |
|
|
|
|
|
|
|
| |
This is useful e.g. on cygwin, where system headers use #pragma directives,
causing annoying false-alarm errors during every g-ir-scanner run.
https://bugzilla.gnome.org/show_bug.cgi?id=630994
|
| |
|
|
|
|
| |
Otherwise we'll bomb out if CFLAGS contains -Werror.
|
|
|
|
|
|
| |
In some cases (such as introspecting a convenience library),
we need to dlsym() modules in the binary we compile, so we
need to add -export-dynamic to the link line.
|
| |
|
|
|
|
| |
We want to use bzip2 since that's what Fedora uses.
|
|
|
|
|
| |
This was broken with the last commit to remove a lot of the special
hacks to link to girepository in the dumper.
|
| |
|
|
|
|
|
|
|
|
| |
This helps us avoid a problematic case where in say jhbuild, using
a system (/usr/lib) glib, adding in -l girepository-1.0 will inject
-L /path/to/builddir, when we don't want that.
https://bugzilla.gnome.org/show_bug.cgi?id=630342
|
|
|
|
| |
Should switch to Makefile.introspection in the future.
|
| |
|
|
|
|
|
|
|
|
|
| |
Kind of silly we weren't doing this before; an empty namespace
is always going to be unintentional.
The "oops I forgot to include .h" files actually happened with
Clutter, but it wasn't caught since the build didn't obviously
fail.
|
| |
|
| |
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=630710
|
| |
|
|
|
|
|
|
|
|
| |
The annotation parser changed to be stricter about annotation content,
but we introduced a regression where a : in the documentation would
wrongly cause the annotation to be skipped.
https://bugzilla.gnome.org/show_bug.cgi?id=630862
|
|
|
|
|
| |
Somehow, this regressed; i have no idea how. We need to be
comparing Include objects, not string versus Include.
|
| |
|
|
|
|
|
|
| |
... because for 'in' arguments only (transfer none) is correct.
https://bugzilla.gnome.org/show_bug.cgi?id=630788
|
|
|
|
|
|
|
| |
Declaring input parameters with transfer other than none is
discouraged, so don't do it in the testsuite.
https://bugzilla.gnome.org/show_bug.cgi?id=630788
|
|
|
|
|
| |
We don't want people making functions which take ownership of
arguments, so don't do it in the test suite.
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=630101
|